Reader Comments

Post a new comment on this article

Food choices and life expectancy: is it just beans?

Posted by Vegard_Lysne on 08 Mar 2022 at 08:50 GMT

Authors and affiliations:
Vegard Lysne1,2,3, Thomas Olsen6, Theogene Habumugisha3,5, Ingunn M.S. Engebretsen3,5, Jutta Dierkes1,2,4

1University of Bergen, MOHN Nutrition research laboratory, and 2Centre for Nutrition, Bergen Norway
3Department of Heart Disease, and 4Department of Laboratory Medicine and Pathology, Haukeland University Hospital, Bergen, Norway
5 Centre for International Health, Department of Global Public Health and Primary Care, University of Bergen, Norway
6University of Oslo, Institute of Basic Medical Science, Oslo, Norway

Preprint: https://osf.io/5wdfp/

The modeling study by Fadnes et al just published in PLOS Medicine has amassed considerable news and social media attention for its claim that an optimized diet could lead to as much as 13 additional years of life expectancy (1). At the time of writing, the study has been viewed more than 120,000 times and generated 247 generally favorable news stories including major news outlets such as CNN and The Telegraph. Although the research effort is ambitious and admirable, the messages that have reached the general public are grossly oversimplified from a nutritional point of view and we have substantial reservations about the methodological approach, interpretations, and conclusions from this study.

The study additively combines effect estimates from published meta-analyses on diet-disease associations for individual foods and food groups to estimate total life-years gained by sustained dietary changes. Although we agree that transitioning to a healthy diet will result in healthy life years gains, we consider the estimates reported in the paper and thereafter in news headlines as implausible for several reasons. Firstly, any changes in energy intake are neglected. This is one of the most important components of dietary exposure and a crucial determinant of health both in under- and overnutrition. Secondly, the effects of changing the intake of individual foods are treated separately and independently, practically ignoring food substitution effects which is a key concept in nutrition research (2) Whenever the intake of one food changes, this is inevitably followed by changes in the intake of other foods and in most cases total energy intake. Thus, health effects must always be interpreted in light of which food is replaced. Thirdly, the individual effect estimates are highly overlapping and influenced by both socioeconomic status (SES) and health consciousness, which are major determinants of longevity. When additively combined, this inevitably causes the model to overestimate the total effect.

We also question the validity of the dietary intake estimates used to calculate the expected effect of specific dietary changes. Food frequency questionnaires (FFQ) are the dietary tool most used for the underlying data. FFQs are mainly used to rank individuals according to dietary exposure, and thus well suited for relative comparisons across different intake levels. However, estimates of absolute intakes should be interpreted with caution due to the presence of systematic measurement error. The model developed by Fadnes et al gives the impression of being able to predict the effect of very specific food intake levels, and we are concerned that this could be interpreted literally by the end-user.

Further, we cannot reasonably assume that the health effects of any food are homogeneous across all population groups, i.e. there is no such thing as ‘one size fits all’ in nutrition. Among other things, the model does not adequately consider acceptability and the situation in low- and middle-income countries (LMICs) or the growing concern of undernutrition among vulnerable subgroups of the population. A fitting example for the issue of LMICs is the PURE study, where diet-disease associations oppose what has been typically observed in Western populations (3) However, baseline nutrition status and food intake in LMICs differ from what is observed in Western affluent populations, and different foods are associated with higher SES.

In contrast to the EAT-Lancet commission’s healthy diet from sustainable food systems (4), the optimal diet described in this paper eliminates red meat and drastically increases legumes and fish above and beyond what is practically feasible or sustainable. Indeed, eliminating intakes of entire food groups could do more harm than good for specific subgroups. For instance, lowering red meat intake would most likely be beneficial for the majority (5,6). At the same time, as a considerable source of protein, energy, and many micronutrients, the effect is likely not the same in populations at risk of undernutrition, such as the elderly, as we previously suggested (7). As emphasized by EAT-Lancet, moderate meat consumption can provide adequate nutrients4, and it can be obtained from sustainably produced livestock (8).

Healthy skepticism has been advised when interpreting observational research on preventive interventions, especially when results appear surprisingly large (9). Unfortunately, we think that the results from Fadnes et al join the ranks of implausible effects. The authors clearly state, and we agree, that the model is not meant for individualized forecasting. Nonetheless, they go on to conclude that clinicians, policymakers, and laypeople could use the calculator to understand the effect of dietary choices. Although we consider the estimated gains in life expectancy as exaggerated, we agree that dietary changes in the general direction described in the paper would benefit the majority in the Western world and would imply reduced meat intake and increased vegetable, legumes, and fish intakes. This includes huge changes for the food systems and questions the sustainability of fish stocks (4). For large parts of the world’s population who have fewer choices improved food security and diversity are a necessity, in combination with a sustainable shift in the diet in the Western world, in order to achieve the sustainable development goal of Zero Hunger.

References:
1 Fadnes LT, Økland J-M, Haaland ØA, Johansson KA. Estimating impact of food choices on life expectancy: A modeling study. PLoS Med 2022; 19: e1003889.
2 Ibsen DB, Laursen ASD, Würtz AML, et al. Food substitution models for nutritional epidemiology. Am J Clin Nutr 2020; published online Dec 9. DOI:10.1093/ajcn/nqaa315.
3 Dehghan M, Mente A, Zhang X, et al. Associations of fats and carbohydrate intake with cardiovascular disease and mortality in 18 countries from five continents (PURE): a prospective cohort study. Lancet (London, England) 2017; 390: 2050–62.
4 Willett W, Rockström J, Loken B, et al. Food in the Anthropocene: the EAT-Lancet Commission on healthy diets from sustainable food systems. Lancet (London, England) 2019; 393: 447–92.
5 Farvid MS, Sidahmed E, Spence ND, Mante Angua K, Rosner BA, Barnett JB. Consumption of red meat and processed meat and cancer incidence: a systematic review and meta-analysis of prospective studies. Eur J Epidemiol 2021; 36: 937–51.
6 Papier K, Fensom GK, Knuppel A, et al. Meat consumption and risk of 25 common conditions: outcome-wide analyses in 475,000 men and women in the UK Biobank study. BMC Med 2021; 19. DOI:10.1186/S12916-021-01922-9.
7 Matre ÅO, Van Parys A, Olsen T, et al. The Association of Meat Intake With All-Cause Mortality and Acute Myocardial Infarction Is Age-Dependent in Patients With Stable Angina Pectoris. Front Nutr 2021; 8. DOI:10.3389/fnut.2021.642612.
8 Röös E, Patel M, Spångberg J, Carlsson G, Rydhmer L. Limiting livestock production to pasture and by-products in a search for sustainable diets. Food Policy 2016; 58: 1–13.
9 Shrank WH, Patrick AR, Brookhart MA. Healthy User and Related Biases in Observational Studies of Preventive Interventions: A Primer for Physicians. J Gen Intern Med 2011; 26: 546.

No competing interests declared.

RE: Food choices and life expectancy: is it just beans?

LarsFadnes replied to Vegard_Lysne on 16 Mar 2022 at 14:08 GMT

Estimating impact of food choices on life expectancy – another step forward but not the final one

Lars T. Fadnes1,2*, Jan-Magnus Økland1,3, Øystein A. Haaland1,3§, Kjell Arne Johansson1,2,3§

We thank Lysne and colleagues for their comments and critical reading of our manuscript[1][2], and will here comment on some of the raised issues.

We agree that several of the news stories linking to our article have focused excessively on point estimates of life expectancy, and our estimates are linked with uncertainties, as the presented uncertainty intervals and sensitivity analyses show.

Lysne et al. comment on energy aspects. It is worth noting that we estimated minor differences in energy between the typical Western and optimal diet approaches, corresponding to a reduction in energy intake of approximately 6%. Thus, energy intake is partly integrated in the model and is also reported in the calculator and manuscript. Further, many of the studies included that contributes to our exposure estimates are adjusted for energy intake, in addition to other factors such as smoking, alcohol intake, and activity. Many of the studies included in the meta-analyses, also adjusted for intake of other food groups. Thus, food substitution effects are to some degree covered. However, it is possible that there is some residual effect overlap between food groups included in our model. Thus, we also presented sensitivity analyses with model adjustments, where the model adjustment of 0.5 accounted for an approximately 50% effect overlap or potential over-estimations of food intake in underlying studies.

We agree that there are several important factors for longevity beyond healthy eating, including socio-economic factors. For people in deep poverty in low-income countries, the model is likely to have limited generalizability. The reason we did not include low-income settings in particular was that we considered the data for these settings to be insufficient and highly uncertain for these exposures. Nevertheless, we acknowledge that many low-income countries have had substantial increase in diabetes type 2, obesity and cardiovascular disease [3, 4], and the eating patterns similar to what we presented as optimal and feasible diet patterns might have a role also in some of these situations [4].

We think the PURE study is interesting [5]. As we read it, the PURE study focuses on macronutrients including fats and carbohydrate, and the study concludes that a very high carbohydrate intake was associated with higher risk of total mortality. Our study focused on food groups and not macronutrients, and we do not consider our results in contrast to the PURE study. For example, carbohydrates include food groups that are generally healthy while eaten in higher quantity, and food groups where research indicate that most people benefit from limiting intake. Examples of the former are fruits and vegetables [6], while sugary-rich food and drinks and refined grains are examples of the latter [7]. Similarly for fat, both nuts and processed meats are often high in fats, although most would agree that nuts are linked with better outcomes in terms of longevity than the latter.

We see the point that a very high intake of fish is linked with some concerns, and some of them were highlighted in our discussion such as content of the toxins (dioxins and polychlorinated biphenyls). The authors stated that the proposed intake of legumes was not feasible or sustainable, but unfortunately, they do not provide references or arguments for this statement.

We agree that estimating exact numbers on health outcomes from eating is a difficult task, particularly on the individual level. Still, we think it is important that integrated evidence is made available, and that we aim for both presenting overall directions of effects as well as estimating effect sizes while transparently presenting the underlying assumptions. We would assume that future research would contribute to more precise estimates and gradually fine-tuning both assumptions and background data. We see our work as a step on the way to gradually more transparent knowledge.

References
1. Lysne V, Olsen T, Habumugisha T, Engebretsen IMS, Dierkes J: Food choices and life expectancy: is it just beans? Comment to "Estimating impact of food choices on life expectancy: A modeling study". In. PLOS Medicine; 2022.
2. Fadnes LT, Økland J-M, Haaland ØA, Johansson KA: Estimating impact of food choices on life expectancy: A modeling study. PLoS Med 2022, 19(1):e1003889.
3. Collaborators GBDCoD: Global, regional, and national age-sex-specific mortality for 282 causes of death in 195 countries and territories, 1980-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet 2018, 392(10159):1736-1788.
4. Collaborators GBDD: Health effects of dietary risks in 195 countries, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet 2019, 393(10184):1958-1972.
5. Dehghan M, Mente A, Zhang X, Swaminathan S, Li W, Mohan V, Iqbal R, Kumar R, Wentzel-Viljoen E, Rosengren A et al: Associations of fats and carbohydrate intake with cardiovascular disease and mortality in 18 countries from five continents (PURE): a prospective cohort study. Lancet 2017, 390(10107):2050-2062.
6. Aune D, Giovannucci E, Boffetta P, Fadnes LT, Keum N, Norat T, Greenwood DC, Riboli E, Vatten LJ, Tonstad S: Fruit and vegetable intake and the risk of cardiovascular disease, total cancer and all-cause mortality-a systematic review and dose-response meta-analysis of prospective studies. Int J Epidemiol 2017, 46(3):1029-1056.
7. Schwingshackl L, Schwedhelm C, Hoffmann G, Lampousi AM, Knuppel S, Iqbal K, Bechthold A, Schlesinger S, Boeing H: Food groups and risk of all-cause mortality: a systematic review and meta-analysis of prospective studies. American Journal of Clinical Nutrition 2017, 105(6):1462-1473.

Affiliations
1 Department of Global Public Health and Primary Care, University of Bergen, Norway
2 Bergen Addiction Research, Department of Addiction Medicine, Haukeland University Hospital, Bergen, Norway
3 Bergen Center for Ethics and Priority Setting, University of Bergen, Norway
§ These authors contributed equally
* lars.fadnes@uib.no

No competing interests declared.